***ALERT NOTE***
SAS Note: SN-003580
Product: ENT MINER
Component: FILTER.OUT
Title:
Filter outliers node creates incorrect data set if 'select all' is used
on the class variables or intervals variables tabs
Text:
In Enterprise Miner, it is common practice to select multiple rows in a
window by click-dragging the mouse, or by using an 'edit - select all'
technique. If you select multiple rows when choosing class variable
values to exclude, or interval variable ranges to include, the node may
not correctly exclude the appropriate observations. If that happens,
then the data sets (and hence the analyses) created by the filter
outlier node are incorrect, even though the node shows that the values
will be excluded. There is no indication that there is any sort of
problem.
The only way to determine if the data sets are incorrect is to examine
the exported data sets by manually verifying that the class variables
contain the correct values, and that the interval variables contain the
correct ranges.
To circumvent the problem, select each row one at a time when setting
values to exclude in the class variables tab, or when setting ranges to
include in the interval variables tab.
EXAMPLE:
Suppose you have a data set with Variables X1 and X2, and both variables
contain a level 'A', among other values.
In the filter outliers node, select the 'class vars tab.' Select both
X1 and X2 either by click-dragging, or by 'Edit' - 'Select All rows.'
Right click in the 'values to exclude column,' 'set values to exclude,'
and click on the histogram bar showing the level 'A'. Close the
histogram. Now the 'values to exclude' column will show that
observations with a value of 'A' for X1 or X2 will be exluded from the
data set. However, due to the bug, level 'A' will be excluded for only
one of the variables and not for the other. Hence the data sets created
contain incorrect observations.
Keywords:
INCORROUT INCORRSTAT FILTER OUTLIER EXCLUDE
Reported Releases:
+------------REPORTED------------+ +------FIXED-----+
System Version Level Version Level
AIX/R 4.0
DigitUnx 4.0
HP800 4.0
Solaris 4.0
WIN/NT 4.0
Win95 4.0
Fixes Available: None
+-------------------------------------------------------------------------
+ NOTE: To unsubscribe you can reply to this mail with:
+ "SIGNOFF tsnews-l"
+ as the only text in the body of the message (without the double quotes).
+ If you have any questions please send them to SUPPORT@SAS.COM
+-------------------------------------------------------------------------
This archive was generated by hypermail 2b29 : Tue Nov 14 2000 - 15:47:31 EST